Miner Face Recognition Based on Improved Singular Value Decomposition

Abstract

This paper depended on that video monitoring and video images in the coal mine were susceptible to dust, light and miner's safety helmet and other special environmental impact. In order to realize the real-time and accurate face recognition rate, and lay good foundation for miners’ behavior characteristics in the subsequent research in intelligent video monitoring coal mine. The inherent degeneration 、stability and Rotational invariance of singular value can reflect the matrix characteristics fully. In face recognition singular value in image matrix as miner characteristics is a very good method. But only it doesn't work using singular value of image as miners face recognition feature, aiming at face recognition, this paper combined partial singular value decomposition and the overall singular value decomposition to suit for miner image face recognition algorithm. This page Improved the previous singular value decomposition algorithm, First The singular value is decomposed in the standard characteristic matrix and projected to get a new algebraic features, Second combined the partial singular value and the overall singular value decomposition to pick up characteristics algorithm, so as to extract and reflect the mine workers face image accurate algebra feature effectively, And by using BP neural network classifier to distinguish recognition, simulation experiments prove that this method of the recognition rate is higher than other methods. The results show that the method combining the part singular value decomposition with the whole singular value decomposition can be faster, more accurate to realize face recognition, and have satisfactory recognition rate. Keywords-Face recognition; Miner face; singular value; video image